Priced per seat.
Usage at cost.

No platform fee, no per-agent tax, no surprise overage invoices. Run 5,000 agent steps free every month and pay only for what production actually consumes.

Developer
$0 forever

For prototypes and single-developer evaluation.

  • 1 seat, 1 environment
  • 5,000 agent steps / month
  • All first-party connectors
  • 7-day trace retention
  • Community support
Start free
Enterprise
Custom

For regulated workloads and dedicated infrastructure.

  • Single-tenant VPC or on-prem
  • Unlimited steps, committed pricing
  • Customer-managed encryption keys
  • Audit export to your SIEM
  • Named solutions architect
  • 99.99% SLA, 24/7 escalation
Talk to sales

Prices in USD. Usage beyond the included step quota is billed at $0.40 per 1,000 steps. Model inference is passed through at provider cost.

Compare

What is in each plan

CapabilityDeveloperTeamEnterprise
Seats included1Unlimited (billed per seat)Custom
Agent steps / month5,000250,000Committed volume
Environments1UnlimitedUnlimited + isolated tenants
Trace retention7 days90 daysUp to 7 years
Eval gates in CI—IncludedIncluded + custom judges
Approval workflows—Slack, Teams, APIAny channel + delegation
SSO / SCIM—SAML, OIDCSAML, OIDC, SCIM, LDAP
DeploymentVeltrix CloudCloud or dedicated VPCVPC or air-gapped on-prem
Data residencyUSUS, EUAny region you operate
SupportCommunityEmail, 1 business day24/7 with named engineer
SLA—99.9%99.99%
BAA / DPA—DPABAA and DPA
Add-ons

Only when you need them

Everything below is optional and priced separately, so the base plan stays predictable.

  • Premium model routing with custom fallback chains
  • Managed evaluation datasets, curated by our research team
  • Dedicated IP egress for allowlisted partner APIs
  • Extended trace retention beyond plan defaults
  • Onboarding sprint: two weeks with a solutions architect

Pricing questions

How are agent steps counted?

One step is a single model turn or tool invocation inside a run. Retries caused by our infrastructure are never billed; retries caused by your tool returning an error are.

Do I pay a markup on model tokens?

No. Inference is passed through at the provider's list price. You can also bring your own provider keys, in which case inference is billed directly by that provider.

Can we start on Team and move to VPC later?

Yes. The API is identical in both modes. Migration is a configuration change plus a data export we run with you; typical cutover is under a week.

Is there a minimum contract?

Team is month-to-month. Annual billing is optional and saves 20%. Enterprise agreements are typically 12 months with a committed step volume.

What happens when we exceed the included steps?

You keep running. Overage is billed at $0.40 per 1,000 steps, and you can set a hard monthly cap that pauses non-critical agents instead of overspending.

Do you offer startup or nonprofit discounts?

Yes — 50% off Team for companies under 20 people and for registered nonprofits, for the first 12 months.

Still comparing?

Send us your current architecture and volume. We will return a written cost model, including what you would spend on building it yourself.

Get a cost model